Multi-criteria Clustering Algorithm for Cooperative Spectrum Sensing

Once the licensed user signal is contaminated by noise uncertainty and shadowing effect occurs, the sensing result from the unlicensed user becomes unreliable. According to cooperative spectrum sensing concept, the reliability of unlicensed users is the major factor that affects the detection performance. In this paper, we propose a new cooperative spectrum sensing solutions - multi-criteria clustering (MCA) - by adopting K mean clustering with the adaptive thresholding scheme. Moreover, once the multi-criteria ranking algorithm is concerned, rank reversal phenomenon where the determined ranking is reversed according to the duplicated version of user performance always occurs and affects the detection performance of cooperative spectrum sensing. Once the unreliable SUs is distinguished from the account, therefore MCA outperforms the detection performance of existing techniques. Moreover, it can avoid the occurrence of rank reversal.